Model for Real-Time Object Searching and Recognizing on Mobile Platform

The method of functional adaptation for objects search and recognition in the video is proposed. This algorithm consists of processing the video image by smoothing and minimization filters which reduces the time of search and recognition of objects. Developed a program to solve the problem of finding and quickly recognizing objects in real time, using Swift language on the iOS mobile platform. A convolutional neural network with YOLOv3 architecture is used. A method of improving the performance of such neural network is proposed, which is based on the quantization of the neural network weights and minimization of the model size and search time of its objects. The frame rate of image processing using the proposed model YOLOv3-KD and models of neural networks type YOLOv3-tiny and YOLOv3-416 are investigated. The possibility of functioning of the proposed approaches in real time is provided.